[thereisnospork]

While I agree with the difficulty to impossibility of finding those specific genes, a whole genotype 'nearest neighbor' or other classification criteria is very interesting:

Considering that both the 'Shaq' phenotype and genotype are known, it wouldn't be too difficult[0] to rank 10,000 embryos per couple in terms of closeness to the 'Shaq' genotype. Then cross reference and weight the 'Shaq-likeness' ranking with the 'Jordan-likeness' ranking and the 'Gretsky-likeness' ranking. To me anyway, that seems like a recipe[1] for, statistically, dramatically improving one's offspring's odds at being a professional basketball player.

[0]Mathematically, anyway

[1]As a counter point, I'd expect to see this sort of thing take off in horse racing if it was productive. Big money and looser ethics.